Online Detection of Water Contamination Under Concept Drift
It addresses public health risks from water contamination by improving detection in water distribution networks, though it appears incremental as it builds on existing anomaly detection techniques.
This study tackled the problem of detecting water contamination in real-time under unreliable chlorine sensors by introducing the Dual-Threshold Anomaly and Drift Detection (AD&DD) method, which effectively identified anomalies and outperformed other methods on two realistic water distribution networks.
Water Distribution Networks (WDNs) are vital infrastructures, and contamination poses serious public health risks. Harmful substances can interact with disinfectants like chlorine, making chlorine monitoring essential for detecting contaminants. However, chlorine sensors often become unreliable and require frequent calibration. This study introduces the Dual-Threshold Anomaly and Drift Detection (AD&DD) method, an unsupervised approach combining a dual-threshold drift detection mechanism with an LSTM-based Variational Autoencoder(LSTM-VAE) for real-time contamination detection. Tested on two realistic WDNs, AD&DD effectively identifies anomalies with sensor offsets as concept drift, and outperforms other methods. A proposed decentralized architecture enables accurate contamination detection and localization by deploying AD&DD on selected nodes.